AI-Powered Adherence Improves Patient Outcomes at Wellth

Wellth provides a digital health platform aimed at improving medication adherence for individuals with chronic conditions. By combining behavioral science with personalized reminders and rewards, the platform helps people maintain consistent adherence to prescribed care plans.

"We founded Wellth with the desire to close the gap between prescribed care and real-world patient outcomes, using time-tested methods and modern technology," explained Matt Loper, Co-Founder and CEO of Wellth. "We aim to help people take control of their health by creating a system that not only reminds them of their care plans, but also provides concrete, verifiable feedback."

How it Works: Reminders and AI-Powered Image Verification

For people on the platform, the Wellth app sends reminders about care tasks and asks users to submit images of their medications or device readings to prove they completed the task. These images are then processed using modern vision AI to confirm the correct medication or reading, which is a critical step in ensuring care plan adherence and accuracy.

This solution has delivered several benefits to people using the platform. Surveys have shown a 16% improvement in medication adherence, a 29% reduction in emergency department admission rate, as well as a 51% lower inpatient visit rate. These stats illustrate the impact of helping people build better habits, enhance outcomes, and avoid unnecessary costs.

A critical component of the platform is machine learning, specifically computer vision. "When considering how to integrate vision AI into our services, finding a secure and robust solution to develop a purpose-built model was top of mind," said Alec Zopf, Co-founder and Chief AI & Automation Officer at Wellth. "We wanted a platform that would allow us to develop models using data we owned, have the ability to scale usage, leverage state-of-the-art AI capabilities, and along the way ensure the privacy of our users’ personal health information. That’s what we found in Roboflow."

Wellth migrated to Roboflow’s platform to develop their proprietary, purpose-built vision models. The platform's end-to-end development tools allowed them to create multiple models to detect several types of medications and health monitoring devices. From there, Wellth was able to deploy their vision models in their application, avoid integrating several disparate machine learning tools, and scale usage seamlessly.

"For a solution like Wellth, focusing our effort on areas that deliver the most value to our users is key," said Zopf. "Roboflow allowed our team to drive model development, without needing to train or hire dedicated machine learning experts, and get computer vision into production faster."

The deployment of their vision AI models has played a significant role in user adherence improvements and benefits. With the ability to automate the verification of care plan tasks, the computer vision tools that Wellth developed on Roboflow have helped users streamline the process of tracking critical healthcare tasks.

The implementation yielded the following key benefits:

• Enhanced User Engagement: By integrating computer vision into the Wellth app, Wellth were able to launch new features supporting patient health like blood pressure meter reading and pill counts, improving outcomes and reducing time spent with manual entry by 90%.
• Scalability: The platform supported the scalability of the vision AI models, allowing Wellth to handle the high volume of daily image submissions: Wellth has scaled from zero to millions of requests seamlessly.
• AI Capability: By leveraging AI-powered labeling, hosted GPU training, and a library of state-of-the-art models, the team at Wellth created and deployed production-ready models in days without the need to hire additional machine learning engineers.
• Data Security: The platform touts HIPAA compliance, ensuring the security and privacy of user data.

"By adopting a comprehensive solution like Roboflow, we got vision AI into production much faster without worrying about stitching together multiple cloud services compared to alternatives," explained Zopf. "This kind of toolset streamlined the development process to create higher quality models and scalable deployments. Importantly, keeping data in a consolidated platform simplifies the process of protecting user data, allowing us to focus on our core mission of improving patient adherence."

‍